
BPS Cognitive Section Conference, Sept 2022
Dr. Christopher J. Wilson
Centre for Applied Psychological Science, Teesside University
Financial distress can increase cognitive load (Mani et al., 2020; Vohs, 2013)
This can affect cognitive processes such as planning, reasoning and decision-making (de Bruijn & Antonides, 2022; Hinson et al., 2003; Hofmann et al., 2012; Mani et al., 2013; Roby & Scott, 2022).
Self-control is a limited cognitive resource and exhausting that resource affects subsequent behaviour (“ego depletion” : Baumeister, 2014; Baumeister et al., 1998; Baumeister et al., 2006, 2008; Baumeister & Vohs, 2018)
Self-control depletion might affect financial or risk-based decisions (Fischer et al., 2012; Gerhardt, 2017; Koppel et al., 2019 )
In the lab, a range of tasks have been used to elicit the effect, including those that target affect, attentional control, thought suppression or response inhibition. It has been replicated across labs (Hagger et al., 2016; Hagger et al., 2010)
Debate about whether this is self-control, or general cognitive fatigue. (Hagger et al., 2010; Inzlicht et al., 2014)
In the current research, inhibitory control tasks are also used (Stroop, Go-noGo)
For the purposes of this research, we will refer use the term Cognitive Exertion
Construal level (Trope & Liberman, 2003) can affect financial decisions (Schmeichel et al., 2011; Ülkümen & Cheema, 2011)
Construal theory:
Construal is both a cause and consequence of cognitive exertion effect (Bruyneel & Dewitte, 2012; Khenfer et al., 2017; Raue et al., 2015; Wan & Agrawal, 2011)
Financial literacy has been shown to increase knowledge and change intentions - but only sometimes examines behaviours (Amagir et al., 2018; Kaiser & Menkhoff, 2020; Mandell & Klein, 2009)
The evidence on efficacy is mixed (Lührmann et al., 2015) and appears to be dependent on many contextual factors (Alessie et al., 2013; Allgood & Walstad, 2016; Chardon et al., 2016; Chen et al., 2018; Foster et al., 2015; Henager & Cude, 2017; Meier & Sprenger, 2013)
Could construal play a role - how information is construed affect subsequent behaviour?
There is research suggesting construal might mediate cognitive exertion effects (Krastev et al., 2020).
However:
Research Questions
Study 1: Following cognitive exertion, how does construal affect financial decision-making?
Study 2: Are there any neurological indicators that distinguish high- and low-construal decisions?
n trials = 20
Expected value of trials calculated as:
value of gain * probability of gain - value of loss * probability of loss
Expected value of trials: 0, 2.5, 5, 7.5, 10

1-way independent, experimental design with 3 conditions.
IV: construal level (High-Construal, Low-Construal and Control condition).
Construal cues were presented as part of the financial decision (coin-toss gamble) task (adapted from Brevers et al., 2018).
DV: Did participants gamble on coin-toss trials? (yes/no)
76 participants took part, 75 completed the study
11 males, 64 females
Random allocation to conditions resulted in the following independent groups:
| Condition | n |
|---|---|
| Control Condition | 26 |
| High Construal | 25 |
| Low Construal | 24 |

Participants are informed that they will be entered into a lottery at the end of the study
If they are randomly selected, one of their coin-tosses will be chosen and they can win the outcome of that specific toss for real
“Your choices in this task do matter”
The model was a significantly better fit than the null model (\(χ^2(4) = 441.47, p < 0.01)\) , Pseudo \(R^2\) (fixed effects) = 0.30
A significant likelihood of not gambling when expected value of the coin-toss was 0 and significant likelihood of gambling in trials with higher expected values than 0 (with the exception of the Expected Value at 2.5)
Model 2 was a significantly better fit than Model 1 \((χ^2(2) = 10.60, p < 0.01)\) , \(\delta\)AIC = -6.6, Pseudo \(R^2\) (fixed effects) = 0.32
Examination of the coefficients showed that both Low Construal and High Construal were significant High Construal Condition (β = -0.97, p < 0.01) compared to the Low Construal Condition (β = -0.49, p < 0.05).
Pairwise comparison of the groups showed that High Construal was significantly different to the Control Condition
There is an effect of construal - High Construal associated with lowest probability of “gambling” and Control condition the highest
Changes to Study 2:
Do the behavioural results replicate Study 1?
Are there different patterns of neurological activation associated with high- and low-construal conditions?
E.g., during the construal cue period immediately preceding the coin-toss decision


Markers sent via serial/usb to fNIRS for specified events
Levels of hbO and hbR are calculated during certain time periods relative to baseline
Current N = 13
No inferential analysis run yet
We can observe fNIRS patterns of activation
Data analysed using a custom R script (will be available on Github)
Preprocessing to remove noise, movement artifacts etc:
.1 Hz Lowpass filter (heartrate, respiration)
Moving average filter (1.5s)
Linear detrending